Active Topics
-
List of well performing apps on Leste (35)
to Maemo 7 / Leste by smatkovi - 2 days, 12 hrs ago -
Porting apps to Leste (43)
to Maemo 7 / Leste by teroyk - 2 days, 19 hrs ago -
Sailfish OS for the Motorola Moto G7 Power (XT1955-5) - (ocean) (13)
to SailfishOS by edp17 - 3 days, 10 hrs ago -
Sailfish OS for the Samsung Galaxy Note 4 (SM-N910C) - (treltexx) (79)
to SailfishOS by edp17 - 3 days, 10 hrs ago -
U-Boot for Nokia RX-51 with BootMenu (updated version 2012.10-rc3-1) (843)
to Alternatives by Arno_11 - 3 days, 16 hrs ago -
Maemo repository (5)
to Community by dredlok706 - 6 days, 11 hrs ago -
Re-commissioning N9 in 2018 (127)
to MeeGo / Harmattan by smatkovi - 6 days, 14 hrs ago - more...
 |
|
2009-06-19
, 12:29
|
|
Posts: 2,535 |
Thanked: 6,681 times |
Joined on Mar 2008
@ UK
|
#2
|
Historical evidence suggests that people will create their own repos if it's easier (one of the reasons why tooling around QA needs to be so good) and that this does cause problems.
In particular, developer A packages a library into their repo because it's needed for their app. Developer B packages the same library and puts it into Extras.
User comes along and has both Extras and dev-A's repo enabled. There are three problem cases:
These aren't theoretical problems, unfortunately - we've seen them in the past.
In particular, developer A packages a library into their repo because it's needed for their app. Developer B packages the same library and puts it into Extras.
User comes along and has both Extras and dev-A's repo enabled. There are three problem cases:
- The two libraries have the same name, but different versions. Both apps require specific naming; but there's been no collaboration or awareness.
- The two libraries have different names but potentially install into the same place. Again, no standards or collaboration on naming.
These aren't theoretical problems, unfortunately - we've seen them in the past.
__________________
Andrew Flegg -- mailto:andrew@bleb.org | http://www.bleb.org
Andrew Flegg -- mailto:andrew@bleb.org | http://www.bleb.org
|
|
2009-06-19
, 12:49
|
|
Posts: 1,418 |
Thanked: 1,541 times |
Joined on Feb 2008
|
#3
|
Originally Posted by jeremiah
You should focus on making it easier to submit applications into Extras. This will make more people do that. The more steps you add to the submission process, the more people will use third-party repos. For example, insistence on auto-builder usage probably makes a lot of people use their own repos.
This negative experience can be avoided if we ask developers to bring their apps "in house" as it were, bring their debs into the official repositories at maemo.org. If we don't do that the feeling is that things will only get worse, more repos will be created and more users will be affected.
I would like the views of the community - what do you think? Should we spend energy and time reducing these repos? Or should we focus on other things?
|
|
2009-06-19
, 13:01
|
|
Posts: 2,535 |
Thanked: 6,681 times |
Joined on Mar 2008
@ UK
|
#4
|
Err, damnit. I said 3 and only listed 2. Can't remember what I had in mind for the third.
__________________
Andrew Flegg -- mailto:andrew@bleb.org | http://www.bleb.org
Andrew Flegg -- mailto:andrew@bleb.org | http://www.bleb.org
 |
|
2009-06-19
, 13:12
|
|
Posts: 1,605 |
Thanked: 1,601 times |
Joined on Mar 2007
@ Southern California
|
#5
|
Originally Posted by Jaffa
And, furthermore...
Err, damnit. I said 3 and only listed 2. Can't remember what I had in mind for the third.
No, really, fms has a good point, of course and streamlining the process of getting an app(s)/library(ies) into Extras-Devel and then Extras should be one of the easiest processes within the maemo.org toolset. This should include an interactive "wizardy" UI as well as automated Q/A (to some degree).
Of course, via mailing list discussion, I know that Jeremiah is working on this sort of thing.
Until then, is there something that can be done to entice developers to go through the current process of getting into Extras? (Unfortunately, I'm not so sure the "make it easier on end-users" argument will work with a lot of people.)
Developers: What would you like to see happen to facilitate this?
Tim
__________________
http://samoff.com
http://samoff.com
 |
|
2009-06-19
, 13:42
|
|
Posts: 170 |
Thanked: 261 times |
Joined on Feb 2009
@ Gothenburg, Sweden
|
#6
|
Originally Posted by fms
So you would like to see a system that does not use autobuilder? But how then would we create the binaries that a user would install? And if you just upload a binary to garage, how can we check that it won't erase the user's hard drive, etc?
You should focus on making it easier to submit applications into Extras. This will make more people do that. The more steps you add to the submission process, the more people will use third-party repos. For example, insistence on auto-builder usage probably makes a lot of people use their own repos.
Are you suggesting we use sbuild and debian native tools or SuSE's Open Build Service?
I don't see any other option other than sending packages through the autobuilder at the moment.
|
|
2009-06-19
, 13:48
|
|
Posts: 1,418 |
Thanked: 1,541 times |
Joined on Feb 2008
|
#7
|
Originally Posted by jeremiah
All I am saying is that autobuilder is a barrier to entry at the moment. The more such barriers you introduce, the more likely it is that a developer will not use Extras.
So you would like to see a system that does not use autobuilder? But how then would we create the binaries that a user would install? And if you just upload a binary to garage, how can we check that it won't erase the user's hard drive, etc?
|
|
2009-06-19
, 13:55
|
|
Posts: 93 |
Thanked: 73 times |
Joined on Sep 2006
|
#8
|
Originally Posted by fms
That's an intersting point of view. I though that since we introduced autobuilder one year ago amount of packages in Extras grew significantly.
All I am saying is that autobuilder is a barrier to entry at the moment. The more such barriers you introduce, the more likely it is that a developer will not use Extras.
| The Following 5 Users Say Thank You to Ed_ For This Useful Post: | ||
|
|
2009-06-19
, 14:43
|
|
Posts: 3,319 |
Thanked: 5,610 times |
Joined on Aug 2008
@ Finland
|
#9
|
I think we can say the opposite, too. Using 3rd party unsupported/unchecked repositories is just that - unsupported and unchecked. Users should be aware of that. IMHO The non-presence of an autobuilder in extras would not lower any barriers, it would just make it more of a dumpster (as then it's every developer for himself) - and then there is no difference if it's just one or a hundred repositories as the contents would get quite messed up over time anyway.
Disclaimer: I'm using launchpad ppa's, pbuilder and similar stuff for quite some time so I might be too easy on the autobuilder approach
Disclaimer: I'm using launchpad ppa's, pbuilder and similar stuff for quite some time so I might be too easy on the autobuilder approach
| The Following 5 Users Say Thank You to attila77 For This Useful Post: | ||
|
|
2009-06-19
, 14:54
|
|
Posts: 3,319 |
Thanked: 5,610 times |
Joined on Aug 2008
@ Finland
|
#10
|
Originally Posted by timsamoff
The most painful part is probably preparing the packages, so a 'packagefy' service comes to mind. A sort of py2deb, only server side - IMHO there is no need for the developer to manually know/create .desktop, .service files, experiment with icon sizes, directory layouts, postinst hooks, etc. Of course, when you create the package in that way once, it's always easier to tweak that than start from zero (even if you have helpful Wiki pages).
Until then, is there something that can be done to entice developers to go through the current process of getting into Extras? (Unfortunately, I'm not so sure the "make it easier on end-users" argument will work with a lot of people.)
Developers: What would you like to see happen to facilitate this?
 |
| Tags |
| maemo repos packages |
«
Previous Thread
|
Next Thread
»
| Thread Tools | |
|
All times are GMT. The time now is 03:29.
I feel that improving the quality control process in the official maemo repos along with other tools to make good software and packaging will mean that users will go to the official maemo.org repositories first to download their software. I also think asking people to take down their repos won't work, for various reasons.
However, others in our community feel differently, and while I won't speak for them, I will try to present their views. Firstly, they see the issue of personal repos as a problem for users since there is no way to do quality assurance on a private repo. If a user downloads something nasty their entire tablet experience is negatively affected.
This negative experience can be avoided if we ask developers to bring their apps "in house" as it were, bring their debs into the official repositories at maemo.org. If we don't do that the feeling is that things will only get worse, more repos will be created and more users will be affected.
I would like the views of the community - what do you think? Should we spend energy and time reducing these repos? Or should we focus on other things?
Thanks!
Last edited by jeremiah; 2009-06-19 at 12:04.